This Program Project will explore theoretical and practical aspects of clinical decision-making, focused on the decision by a physician to order a diagnostic test or procedure for evaluation of a diagnostic hypothesis. A multidisciplinary research and development team has been assembled to pursue four aspects of this problem: (1) We will develop clinical decision support programs to aid the referring physician in the quantitative assessment of probabilities implicit in the decision to order a test, and seek to demonstrate a beneficial effort of such assistance on the utilization of tests. (2) We will develop data bases to identify potential areas of high yield for such consultations, and to characterize the "diagnosticity" of the tests involved. (3) An artificial intelligence model of decision-making will be implemented which uses the concept of "threshold" for diagnosis to link probabilistic data to actions. (4) Theoretical investigations of the concept of threshold will be pursued, aimed at characterizing the factor involved.